Comparison of various statistics for identification of maize inbreds with favourable alleles useful for the improvement of single cross hybrid in maize (Zea mays L.)

Theory and methods for identifying inbred lines (P_W) where favourable dominant alleles are not present in an elite single cross hybrid (P₁ × P₂) have recently been developed and gaining popularity in modern maize breeding programmes. In the present study, five methods are used to estimate the number of favourable alleles present in an inbred but not present in an elite hybrid namely estimator of Dudley (μ G′), upper bound estimates (UBND), predicted three way cross (PTC), net improvement statistic (NI) and unbiased estimator based on combining ability model (μa[B + G]) and were applied to a maize 10 x 10 diallel cross data. Besides these estimates, general combining ability estimates and mean per se value of the donor inbreds were also used for comparison studies. Results revealed that, though each and every method has its own conceptual and computational advantages as well as limitations, μ G′, NI, PTC and UBND, have higher inter se correlations. Hence, it is indicated that anyone of these methods can successfully select the donor with the highest favourable alleles for the improvement of single cross hybrid. In general, no association was found between mean per se and all other estimates for all the three characters under consideration. Similarly, GCA and (μa [B + G]) failed to exhibit closer association in grain yield per plant and ear length.